Two people have eleven meeting hours this week.
The first has them in four blocks: Monday afternoon, Wednesday morning, Wednesday afternoon, Friday morning. Three of her days contain a clear stretch of four hours or more.
The second has them spread across ten sessions, roughly one every ninety minutes, every day. His longest uninterrupted stretch all week is fifty-five minutes, and it is on Friday at 4pm.
Same load. The second person is drowning and the first one is not, and if he audits his calendar looking for meetings to cut he will find nothing obviously wrong. That is calendar fragmentation: it is invisible to every metric that counts hours, which is every metric most people compute.
What fragmentation actually costs
Two things, and they compound.
The restart. Coming back to complicated work is not free. You have to reload what you were doing, where you had got to, and what you had decided not to do — and the reload is longer the more of the thing you were holding in your head. Nobody can tell you the size of that cost in minutes for your work specifically. Figures do circulate; the widely repeated one is not from a source I have been able to verify directly, and I am not going to hand you a number I have not read the paper for. Measure your own version instead: for a week, note the clock time when you sit down to a task and the clock time you actually start producing. The gap is your restart cost, and it is real whatever anyone’s average says.
The shadow. The forty minutes before a meeting are not forty minutes. You know something is coming, so you do not start anything that could get cut off, so you do the small thing instead — inbox, approvals, a review that fits in the window. The meeting has claimed time it does not appear to occupy. This is why a day with two well-spaced meetings can produce nothing: those two meetings cast four shadows between them.
Add them together and the arithmetic is unkind. A 30-minute meeting placed in the middle of a two-hour window does not cost 30 minutes. It costs the 30, plus the shadow before it, plus the restart after, and it converts one usable block into two unusable ones.
Measuring calendar fragmentation
You need three numbers. All of them come out of the same four-week export you pulled for the calendar audit.
1. Median longest block. For each working day, find the longest continuous stretch with no meeting. Treat any gap under 30 minutes as no gap at all — a 25-minute window between two calls is a waiting room. Take the median across all your days.
Median, not mean. One empty Friday pulls a mean up by an hour and hides four broken days behind it.
2. Switch count. Count the number of times per day you enter or leave a meeting. A day with six 30-minute meetings has twelve transitions; a day with one three-hour workshop has two. Same hours, wildly different days. Take the average across your weeks.
3. Contiguity ratio. Total meeting hours ÷ number of separate meeting sessions, where back-to-back meetings with no gap count as one session. This gives you the average length of an uninterrupted meeting run.
A high contiguity ratio is good. It means your meetings are stacked, which means the space between them is real. A ratio near your average meeting length means every meeting is standing alone in its own patch of ruined afternoon.
Then read the three together:
| Median block | Switch count | What you have |
|---|---|---|
| 3h+ | Under 6 | A well-shaped week. If it still feels bad, look at load, not shape. |
| 1.5–3h | 6–8 | Workable for most roles. Fine for responsive work, tight for deep work. |
| Under 90 min | 8+ | Fragmented. Declining meetings will not fix this — moving them will. |
| Under 60 min | 10+ | Nothing that requires sustained attention is getting done, regardless of total load. |
The thresholds are not research; they are a reading aid. What matters is the direction of travel between one quarter and the next, and whether the median block clears the length your particular work needs.
Why the usual advice does not fix it
“Decline more meetings.” Cutting three of your ten sessions leaves seven, still spread across five days. Your load drops 30% and your median block barely moves. Declining is the right tool for load and the wrong one for shape.
“Block focus time.” A block placed in a gap that was already too short to use is not focus time, it is a label on a waiting room. Blocks work only after the meetings around them have been moved, which is the step people skip.
“Use an AI scheduling tool.” Some of them do defragment well; that is genuinely what they are for. But they can only move what is movable, and if your fixed points are five recurring meetings owned by five different people, the tool will report that it has nothing to work with. Fix the fixed points first, then consider a tool, in that order.
What actually fixes it
Fragmentation is a placement problem, so every real fix moves something.
Move one recurring meeting. This is the highest-return action available to you and almost nobody does it, because moving feels like a bigger ask than declining. It is a much smaller one. You are not questioning whether the meeting should exist; you are asking for a different slot. Organisers say yes to this far more readily than they say yes to anything else.
Pick the meeting that most often sits alone in the middle of a morning, and ask for it to move next to another one:
“Would it work to move this to 10:30? I’ve got the platform sync at 10:00 and it would save me breaking up the morning twice. Happy either way if 9:30 suits people better.”
Two consolidated meetings and a clear afternoon beats two isolated meetings and no afternoon. That is the entire manoeuvre.
Declare a direction. Ask people to book you in the afternoon. Not a policy, not an announcement — a line in your booking link’s availability, a sentence in your team channel, a habit of proposing afternoon slots when you are the one suggesting times. Half of your meetings are scheduled by people who have no preference and are picking the first open slot they see.
Set your working hours so the first slot is not 9am. Both Google Calendar and Outlook let you define working hours, and most people booking you will respect what the grid shows. If the first bookable slot is 11:00, your mornings survive by default rather than by defence.
Use the 30-minute gap rule when you accept. Before accepting an invite, look at what it does to the day. If it converts one two-hour block into two 45-minute ones, propose a different time in the acceptance rather than accepting and resenting it. The moment of acceptance is when you have the most leverage and the least friction, and it is the moment everyone lets pass.
Stack, do not spread. When you are choosing between two open slots, take the one adjacent to an existing meeting every time. It feels worse — three meetings in a row is unpleasant — and it produces a better week. The unpleasantness is concentrated instead of distributed, which is the trade you want.
The one that needs your manager
If your fragmentation comes from meetings your manager schedules or requires, you cannot fix it from below and you should stop trying. Bring the three numbers to a 1:1 and ask for a shape rather than a reduction:
“My meeting hours are fine — it’s about eleven, which is what the role needs. The problem is that they’re spread across every day, so the longest clear stretch I get is under an hour and the forecast work isn’t getting done. Could we look at putting the team meetings on Tuesday and Thursday? Same meetings, same hours, different days.”
That ask is easy to grant because it concedes nothing. Nobody has to admit a meeting is unnecessary, no one loses access to you, and the total stays the same. Requests that cost the other person nothing get approved at a rate that requests for reduction never do — which is also why the decline that doesn't say no works when a straight decline would not.
Common questions
Is fragmentation worse than a heavy meeting load? For work that requires sustained attention, usually yes, and it is much harder to argue about because the hours look reasonable. For responsive work — support, coordination, most gatekeeping — a fragmented day may be exactly the correct shape, and a long uninterrupted block would be the anomaly.
What if my role has no long tasks? Then a low median block is not a problem. Do not fix a number that is not costing you anything. Check the switch count instead: even without deep work, twelve context switches a day has a fatigue cost that shows up by Thursday.
Should I batch all my meetings into two days? It works well and it costs something specific: on the two heavy days you are unreachable and useless for anything else, and colleagues learn it. Worth doing if your work has genuinely long tasks; a bad trade if you are the person people need to unblock them.
My meetings are all 30 minutes. Should I make them shorter? Shorter meetings make fragmentation worse, not better. Six 15-minute meetings scattered across a day cost more than three 30-minute ones stacked. Duration is not the variable that matters here. Adjacency is.